dstack to manage clusters of on-prem servers for AI workloads with ease
ā¢
7
chansung park PRO
chansung
AI & ML interests
None yet
Recent Activity
reacted
to
their
post
with š
1 day ago
Simple summary on DeepSeek AI's Janus-Pro: A fresh take on multimodal AI!
It builds on its predecessor, Janus, by tweaking the training methodology rather than the model architecture. The result? Improved performance in understanding and generating multimodal data.
Janus-Pro uses a three-stage training strategy, similar to Janus, but with key modifications:
ā¦ Stage 1 & 2: Focus on separate training for specific objectives, rather than mixing data.
ā¦ Stage 3: Fine-tuning with a careful balance of multimodal data.
Benchmarks show Janus-Pro holds its own against specialized models like TokenFlow XL and MetaMorph, and other multimodal models like SD3 Medium and DALL-E 3.
The main limitation? Low image resolution (384x384). However, this seems like a strategic choice to focus on establishing a solid "recipe" for multimodal models. Future work will likely leverage this recipe and increased computing power to achieve higher resolutions.
posted
an
update
1 day ago
Simple summary on DeepSeek AI's Janus-Pro: A fresh take on multimodal AI!
It builds on its predecessor, Janus, by tweaking the training methodology rather than the model architecture. The result? Improved performance in understanding and generating multimodal data.
Janus-Pro uses a three-stage training strategy, similar to Janus, but with key modifications:
ā¦ Stage 1 & 2: Focus on separate training for specific objectives, rather than mixing data.
ā¦ Stage 3: Fine-tuning with a careful balance of multimodal data.
Benchmarks show Janus-Pro holds its own against specialized models like TokenFlow XL and MetaMorph, and other multimodal models like SD3 Medium and DALL-E 3.
The main limitation? Low image resolution (384x384). However, this seems like a strategic choice to focus on establishing a solid "recipe" for multimodal models. Future work will likely leverage this recipe and increased computing power to achieve higher resolutions.
upvoted
an
article
1 day ago
Open-R1: a fully open reproduction of DeepSeek-R1
Articles
Organizations
chansung's activity
upvoted
an
article
1 day ago
Article
Open-R1: a fully open reproduction of DeepSeek-R1
ā¢
374
upvoted
an
article
10 days ago
Article
Introducing multi-backends (TRT-LLM, vLLM) support for Text Generation Inference
ā¢
61
upvoted
a
paper
about 2 months ago
upvoted
an
article
3 months ago
Article
Llama 3.2 in Keras
ā¢
12
upvoted
an
article
4 months ago
Article
dstack to manage clusters of on-prem servers for AI workloads with ease
By
ā¢
ā¢
7
upvoted
a
collection
5 months ago
upvoted
an
article
5 months ago
Article
Efficient Deep Learning: A Comprehensive Overview of Optimization Techniques š š
By
ā¢
ā¢
48
upvoted
a
paper
5 months ago
upvoted
an
article
5 months ago
Article
dstack: Your LLM Launchpad - From Fine-Tuning to Serving, Simplified
By
ā¢
ā¢
12
upvoted
an
article
6 months ago
Article
The Workflow of PEFT
By
ā¢
ā¢
19
upvoted
3
articles
9 months ago
Article
Deploying š¤ ViT on Vertex AI
ā¢
2
Article
Expanding Model Context and Creating Chat Models with a Single Click
By
ā¢
ā¢
37Article
Faster fine-tuning using TRL & Unsloth
ā¢
47
upvoted
an
article
10 months ago
Article
CodeGemma - an official Google release for code LLMs
ā¢
100